tom

CVEN4405: Human Factors in Civil and Transport Engineering

Automated and Connected Transport Systems 1: Overview

Term 3, 2020

Week 10, Lecture 1a

1

CVEN 4405 Human Factors in Civil and Transport Engineering

Term 3 2020 Week 10 - Lecture 1a

PRESS RECORD BUTTON

+

SHARE SCREEN

2

Lecture Recordings

PLEASE NOTE.

All lectures today are being recorded.

Participation in this meeting indicates your consent to be included in the meeting recording.

3

Welcome Back!

4

Course Coordinator and Lecturer

Prof. Michael Regan, PhD Professor of Human Factors

Research Centre for Integrated Transport Innovation (rCITI) School of Civil and Environmental Engineering

University of NSW Sydney

T: +61 (0)2 9385 9504 E: [email protected]

Staff Webpage

5

The CVEN 4405 Teaching Team

Coordinator and Lecturer Prof. Michael Regan Professor of Human Factors Research Centre for Integrated Transport, UNSW Sydney E: [email protected]

Teaching Fellow Dr Prasannah Prabhakharan Research Fellow Research Centre for Integrated Transport, UNSW Sydney E: [email protected]

Demonstrator Mitch Cunningham E:[email protected]. au

6

Review of Last Lecture

Mr David McTiernan Office of the National Rail Safety Regulator

“The Safe System Approach to Road Transport”

7

This lecture - Overview

• Automation • Automated vehicles • Connected vehicles • Implications for Road and Traffic Engineering

8

Learning Outcomes

CLO1: Explain the fundamental principles of HF that can be used by civil and transport engineers to facilitate user- centred design

CLO2: Apply HF principles, methods and data to the design of road and traffic management systems

CLO3: Plan for the integration of HF into the design lifecycle of the road and traffic management system

9

Introduction

In this final day of the course, we’ll learn about how the world of transport is changing as:

– vehicles become increasingly capable of automating many traditional driving tasks; and

– vehicles become increasingly connected with each other, with road infrastructure, with service providers and with other “things”

We’ll start off with a short video to introduce you to this new world of transport

Introduction to Automation

11

Automation: Why Automate?

• Discussion: Why would we want to automate anything?

12

Automation: Why Automate? (2)

• When it is dangerous for humans to perform the same tasks – e.g. robots locating land mines

• When it is impossible for humans to perform tasks e.g. aids for the disabled

• When tasks are difficult e.g. autopilots in aircraft • To extend human capability e.g. forward collision warning • Just because we can – even though it may provide little or

no value to the human user.

Source: Wickens et al (2004)

13

Automation: Levels of Automation

From Complete Manual Control to Complete Automatic Control:

• Automation offers no aid to driver • Automation suggests multiple alternatives and highlights best one e.g. nav

system • Automation selects an alternative and suggests to human e.g. nav system • Automation carries out an action if human approves • Automation provides human with time to veto action before carrying out action • Automation carries out action then informs human • Automation carries out action and informs person only if asked • Automation carries out action and ignores the driver

Source: Wickens et al (2004)

Automated Vehicles

15

Automated Vehicles: Definitions

Automated vehicles – are those which use electronic or mechanical devices to replace human driving functions

Autonomous vehicles – are those which use electronic or mechanical devices to replace all human driving tasks:

• They can drive themselves – they are independent and self sufficient. • They do not need to cooperate with other vehicles or infrastructure

Source: Shladover (2017), p. 190

16

Automated & Connected Vehicles: Video

Source: https://www.youtube.com/watch?v=1lUkyCYdAEY (5:32)

17

Automated Vehicles: SAE Levels

Source: https://www.sae.org/

18

Automated Vehicles: SAE Levels - Examples

Source: Shladover (2017) p. 194

Level Example Systems Who is in Control?

1 Adaptive cruise control or lane keeping assist Driver 2 Adaptive cruise control and lane keeping assist

(operating together) Driver

3 Traffic Jam Assist for freeway (e.g. Mercedes, Tesla, Volvo etc)

Vehicle – but driver has to intervene if necessary

4 • Highway Driving Pilot • Driverless Shuttles • Driverless valet parking in garage • Automated buses/trucks on special transit

lanes • Passenger cars on limited access freeways

Vehicle

5 Robo taxis Vehicle

19

Automated Vehicles: Automating Traditional Driving Activities Brown (1986) describes driving as involving 6 functional

activities 1. route finding 2. route following 3. lateral and velocity control 4. collision avoidance 5. rule compliance 6. vehicle monitoring

Source: Brown (1986)

20

Automated Vehicles: Potential Benefits

Improved safety Improved mobility Reduced congestion Improved productivity Reduced fuel consumption Cleaner environment

Connected Vehicles

22

Connected Vehicles: The Future

The transport system of the near future will be one in which vehicles, driven manually and autonomously, become increasingly connected with the rest of the road and traffic system, and with the rest of the world.

Automated vehicles are already to some extent “connected” with the road and traffic system.

• E.g. Lane departure warning/Lane Keeping Assist systems have on- board cameras that “read” line markings to determine whether drivers are driving within their lanes.

23

Connected Vehicles: Types of Connections

As we saw in the video, vehicles will become increasingly connected, and there are several ways in which this may happen:

• V2V (vehicle to vehicle) • I2V (infrastructure to vehicle) • V2I (vehicle to infrastructure) • V2P (vehicle to pedestrian) • V2X (vehicle-to-anything)

Source: Shladover (2017), p. 196

24

Connected Vehicles: V2V

Vehicle-to-vehicle (V2V) connectivity will enable applications such as: • “cooperative collision warnings and hazard alerts; • cooperative collision mitigation or avoidance

incorporating active braking; • cooperative adaptive cruise control, with tighter vehicle-

following control • close-formation automated platooning • automated manoeuvre negotiation at merging locations

or intersections”

Source: Shladover (2017), p. 196

25

Connected Vehicles: I2V

Infrastructure-to-vehicle (I2V) connectivity will enable applications such as:

• “providing traffic signal status information in real time for in-vehicle display, signal violation warning, or green wave speed advisories to drivers;

• providing traffic and weather condition information and real-time routing advisories to drivers

• fleet management functions of vehicle routing and scheduling; • access control to closed facilities; • variable speed limits and advisories provided directly to drivers or their vehicles

(I2V cooperative adaptive cruise control); • end of queue warnings • active support for lane guidance.”

Source: Shladover (2017), p. 192

26

Connected Vehicles: V2I

Vehicle-to-Infrastructure (V2I) connectivity can enable applications such as:

• “vehicle probe data applications providing detailed traffic information (speed, volume, travel time, queue length, and stops) or road surface condition information (pavement roughness or slippery conditions);

• mayday and concierge services (such as OnStar); electronic toll collection and parking payments;

• traffic signal priority requests; • vehicle status information for fleet management (especially for transit

and trucking fleets).”

Source: Shladover (2017), p. 193

27

Connected Vehicles: V2P

Vehicle-to-Pedestrian (V2P) connectivity: • “The V2P category can be considered to include any

vulnerable road user, including pedal cyclists, who may be carrying a nomadic device that can communicate with nearby vehicles.”

Source: Shladover (2017), p. 192

28

Connected Vehicles: V2X

Vehicle-to-Anything (V2X) connectivity: – “V2X stands for vehicle-to-anything - analogous to the

current interest in the “internet of things” in which virtually every device could be connected to any other device.”

Source: Shladover (2017), p. 192

29

Connected Vehicles: Wireless Technologies

Connected vehicles can rely on a variety of wireless communication technologies for their connectivity:

• 5.9 GHz DSRC: a special WiFi-like technology designed for road transportation applications

• WiFi • Cellular communications: e.g. 4G LTE and WiMAX technologies and 5G

cellular (when available) • Satellite communication systems e.g. for remote areas without cellular

service • Bluetooth: providing very short-range and low-bandwidth service to

support some applications

Source: Shladover (2017), p. 192

30

Connected Vehicles: Implications

This increased connectivity will have implications for road design and traffic management going into the future.

For example, to support I2V functionality of highly-automated vehicles, road infrastructure will need to become “smart”.

I2V communication has potential to replace some of the functions currently performed by road signs and signals • e.g. road-related information could be communicated to the vehicle

and driver wirelessly rather than through traditional visual means (e.g., through road signs and signals)

Source: CAR (2017), Regan, (2004a, 2004b); Centre for Automotive Research (2017).

31

Connected Vehicles: Implications (2)

Traffic signs of the future, for example, may not necessarily be static or dynamic road signs attached to poles on the side of the road.

They may be ‘located’ inside the vehicle in the form of text or graphic images on heads-down or heads-up displays, or as auditory displays.

For the driver, the distinction between traffic information presented inside and outside the vehicle will become increasingly blurred.

It will become increasingly important for vehicle designers to ensure that traffic-related information conveyed to the driver within the vehicle does not distract, overload or confuse the driver

Source: Regan, (2004b)

32

Connected Vehicles: Implications (3)

V2V connectivity will enable vehicles to constantly talk to each other, allowing smoother traffic by optimising speeds and allowing closer safe following distances (e.g., ‘platooning’); and, therefore, increasing traffic throughput

This could have implications for the number of lanes required to move traffic efficiently; less might be needed if V2V allowed for increased traffic throughput

Source: CAR (2017), Cunningham, Regan & Cairney (2017).

33

Connected Vehicles: Implications (4)

Lanes dedicated to CAVs would not require additional width to accommodate for human error.

So, lane width could be closer to actual vehicle width, and be reduced by as much as 20 percent if vehicle dimensions remain roughly constant.

These are just a few implications of connectivity for road and traffic engineering.

Source: Federal Highway Administration, (2015)

34

Connected Vehicles: Challenges

Shladover (2017) identifies four key challenges for transport agencies in deploying CV technologies:

1. Funding: existing resources are scarce and maintenance costs for CV technologies are expected to be high.

2. Staffing/Workforce: Information technology systems such as CV will require a different mix of skills

Source: Shladover (2017), p. 193

35

Connected Vehicles: Challenges (2)

3. Understanding likely impacts: the costs and benefits of CV systems are more difficult to estimate/understand than for traditional transportation technologies

4. Inconsistent deployment: If some jurisdictions implement cooperative infrastructure while others don’t’ – then a CV system that depends on this infrastructure will frustrate travellers

• e.g. electronic tolling

Source: Shladover (2017), p. 193

36

Connected & Automated Vehicles: Synergies

Connected and automated vehicles provide complementary contributions to improving performance and safety of the transport system

CV technology provides additional data that can be provided to AV systems to improve their performance and safety:

• e.g. information that AV sensors cannot see (e.g. black ice on road) • e.g. information about vehicles beyond the immediate line of sight that

cannot be sensed by AVs (which can reduce shock waves) • e.g. information about pedestrians and other road users that cannot be

sensed by AVs

Source: Shladover (2017), p. 197

37

Summary

In this lecture we have learnt about how the world of transport is changing as: – vehicles become increasingly capable of automating many

of Brown’s (1986) driving tasks; and – vehicles become increasingly connected with each other,

with road infrastructure, with service providers and with other “things”.

Synergies between AVs and CV technology were discussed, along with some implications and challenges for road and traffic engineering.

38

Self-Directed Reading

Shladover, S.E. (2017). Connected and automated vehicle systems: Introduction and overview. Journal of Intelligent Transportation Systems, 22:3, 190-200.

https://doi.org/10.1080/15472450.2017.1336053

39

Questions ?

Over to ….

Prof. Michael Regan, PhD Research Centre for Integrated Transport Innovation

(rCITI) Room 112, Civil Engineering Building (H20)

E: [email protected]

CVEN4405: Human Factors in Civil and Transport Engineering

Automated and Connected Transport Systems 2: The

Changing Role of the Driver Term 3, 2020

Week 10, Lecture 1b

42

CVEN 4405 Human Factors in Civil and Transport Engineering

Term 3 2020 Week 10 - Lecture 1b

PRESS RECORD BUTTON

+

SHARE SCREEN

43

Lecture Recordings

PLEASE NOTE.

All lectures today are being recorded.

Participation in this meeting indicates your consent to be included in the meeting recording.

44

Welcome Back!

45

Course Coordinator and Lecturer

Prof. Michael Regan, PhD Professor of Human Factors

Research Centre for Integrated Transport Innovation (rCITI) School of Civil and Environmental Engineering

University of NSW Sydney

T: +61 (0)2 9385 9504 E: [email protected]

Staff Webpage

46

The CVEN 4405 Teaching Team

Coordinator and Lecturer Prof. Michael Regan Professor of Human Factors Research Centre for Integrated Transport, UNSW Sydney E: [email protected]

Teaching Fellow Dr Prasannah Prabhakharan Research Fellow Research Centre for Integrated Transport, UNSW Sydney E: [email protected]

Demonstrator Mitch Cunningham E:[email protected]. au

Review of Last Lecture

• Automation • Automated vehicles • Connected vehicles • Implications for Road and Traffic Engineering

This Lecture - Overview

• Impact of Automated and Connected Vehicles on Driving

• Impact of automation on Brown’s driving functions

• The Changing Role of the Driver

• Knowledge, Skills and Behaviours required to use Automated Vehicles

• Implications for Road and Traffic Engineering

• Technical Challenges in Automating Brown’s driving functions

49

Learning Outcomes

CLO1: Explain the fundamental principles of HF that can be used by civil and transport engineers to facilitate user- centred design

CLO2: Apply HF principles, methods and data to the design of road and traffic management systems

CLO3: Plan for the integration of HF into the design lifecycle of the road and traffic management system

Impact of Automated and Connected Vehicles on Driving

• As motor vehicles become increasingly equipped with automation, and connectivity, the driving functions and tasks performed by drivers will change.

• This will change the repertoire of knowledge, skills and behaviours required by drivers to safely operate these vehicles.

Impact of Automated and Connected Vehicles on Driving

• In this lecture, we’ll discuss how Brown’s (1986) driving functions will change with increasing automation.

• As road traffic engineers, you need to be thinking about what driving functions you will need to continue to support drivers to perform safely, and what new functions you will need to support as the drivers role changes.

The Driving Task - Brown’s Model Revisited

Brown (1986) characterises driving as involving six functions:

• Route finding • Route following • Velocity control • Collision avoidance • Rule Compliance • Vehicle monitoring

Impact on Brown’s driving functions - SAE Levels 0, 1 and 2

• Route Finding – automated by satellite navigation (Level 1)

• Route Following – partial automation by Lane Keeping Assist (Level 1)

• Velocity Control – partial automation by: • Adaptive Cruise Control (Level 1) • Adaptive Cruise Control with Lane-centring (Level 2)

Impact on Brown’s driving functions - SAE Levels 0, 1 and 2 (3)

• Collision avoidance: partially automated by:

• Antilock Brakes (Level 0) • Electronic Stability Control (Level 0) • Autonomous Emergency Braking (AEB; Level 0) • Lane Keeping Assist (Level 1) • ACC with Lane-centring (Level 2)

Impact on Brown’s driving functions - SAE Levels 0, 1 and 2 (4)

Rule Compliance: partially automated by: • Intelligent Speed Adaptation (Level 1) • Top Speed Limiters (Level 1) • Convention Cruise Control (Level 1) • Adaptive Cruise Control (Level 1)

Vehicle monitoring: • the requirement for Vehicle Monitoring will increase, going from

SAE Levels 0 to 2, as more ADAS technologies issue an increasing number of warnings, and there is a greater requirement for the driver to monitor the status of ADAS and automated driving features.

Impact on Brown’s driving functions - SAE Levels 0, 1 and 2 (5)

Vehicle monitoring: • The requirement for Vehicle Monitoring will increase,

going from SAE Levels 0 to 2 • More ADAS technologies will issue an increasing

number of warnings • There is a greater requirement for the driver to monitor

the status of ADAS and automated driving features.

Impact of Impact on Brown’s driving functions - SAE Levels 3, 4 and 5

• At Level 3 and beyond, the vehicle is considered to be driving when automated driving features are engaged.

• However, at Level 3, the driver must drive the vehicle when requested by the vehicle.

• At Levels 4 and 5, no such requests are made by the vehicle.

Impact of Impact on Brown’s driving functions - SAE Levels 3, 4 and 5 (2)

• At Level 3, in terms of Brown’s (1986) functional model of driving, automated driving features will automate almost all the functional activities described in the model, but only under specific limited conditions e.g. on freeways.

• At Level 3 there remains a requirement for Vehicle Monitoring – for the driver to be vigilant for takeover requests and be ready to regain control of the vehicle and resume manual driving.

Impact of Impact on Brown’s driving functions - SAE Levels 3, 4 and 5 (3)

At Level 4, in terms of Brown’s (1986) functional model of driving, automated driving features will automate all the functional activities described in the model but, again, only under specific limited conditions e.g. on freeways.

At SAE Level 4 there is no requirement for Vehicle Monitoring, as the vehicle can bring itself to a safe operating condition if the vehicle cannot continue to drive autonomously.

The driver has no functions to perform.

Impact of Impact on Brown’s driving functions - SAE Levels 3, 4 and 5 (4)

At Level 5, in terms of Brown’s (1986) functional model of driving, automated driving features will automate all the functional activities described in the model in all conditions.

The driver has no functions to perform.

The Changing Role of the Driver

• With increasing automation, the primary function will no longer be active control of the vehicle, but supervisory control (Sheridan, 2002), especially for Level 3 vehicles in the SAE (2018) taxonomy which requires the driver to regain control of the vehicle when requested by the vehicle.

• In highly and fully automated vehicles (SAE Levels 4 and 5 vehicles), humans are likely to play a passive observer role, as they will not be required to provide any input into the driving task (Spulber, 2016).

62

The Changing Role of the Driver (2)

Source: Spulber (2016)

SAE Level Level 0 Level 1 Level 2 Level 3 Level 4 Level 5

Type of driver support/automa ted driving

No automation

Driver assistance

Partial automation

Conditional automation

High automation

Full automation

Role of human operator

Active controller

Active controller

Active controller Supervisor

Passive observer

Passive observer

Knowledge, Skills and Behaviours required to use Automated Vehicles

• With these changing roles, will come a change in the knowledge, skills and behaviours that are required to safely operate automated vehicles from Levels 1 to 3.

Source: Regan, Prabhakharan et al. (2019)

KnowledgeFor example:

• Environmental conditions that may degrade system performance

• Environmental conditions that are unsuitable for the use of system

• System performance limitations not related to driving conditions, including causes of degraded system performance

• Symptoms of system malfunctions

Source: Regan, Prabhakharan et al. (2019)

Skills For example:

• Recognizing environmental conditions that may cause system performance to be ineffective or degraded

• Recognizing environmental conditions that may degrade system performance

• Distinguishing between normal and aberrant system performance

Source: Regan, Prabhakharan et al. (2019)

Behaviours

For example:

• Maintaining vigilance of the driving environment, system operating modes and associated vehicle performance

• Willingness to use system when appropriate to do so • Avoidance of the use of the system when inappropriate to

do so.

Source: Regan, Prabhakharan et al. (2019)

Implications for Road and Traffic Engineering

• For as long as there remains a requirement for the driver to perform some or all of Brown’s (1986) driving functions – at SAE Levels 0 to 3 – the capabilities and limitations of drivers to perform these tasks will need to be taken into account in design the road and traffic system.

• When the the driver is no longer required to perform Brown’s (1986) driving functions – SAE Levels 4 and 5 – the role of the road and traffic engineer will be to design the road and traffic environment in a way that supports automated vehicles to perform these tasks

•

Implications for Road and Traffic Engineering (2)

Shladover (2017, p. 197) points to the important role that transport infrastructure will need to play in future as vehicles

become highly automated and autonomous:

“The bulk of the attention in CV and AV research until now has been focused on the vehicles, with the infrastructure

side being cast in the supporting role. However, in order to implement I2V or V2I applications, the “I” as in infrastructure

is an essential element. It will also be an important supporting element for the security aspects of V2V

communications and applications.”

Implications for Road and Traffic Engineering (3)

Shladover (2017) identifies a number of topics that transportation agencies will need to be well-informed about in order to make intelligent decisions about CV and AV deployment and operation, including: • “Guidance about modifications that may be needed to

standard practices in transportation design and operations to facilitate use of CV and AV technologies (in areas such as roadway geometric design, signage and markings, and data management).”

Source: Shladover (2017, p. 198)

Technical Challenges in Automating Brown’s driving functions

A quote from Shladover (2017):

“ Although no explicit safety standards have been specified yet, it is not unreasonable to expect the automated driving system to maintain at least the level of safety of average

human drivers today. As explained in Shladover (2014), for road travel in the US today this represents a mean time

between fatal crashes of more than 3 million vehicle hours of driving and a mean time between injury crashes of about

65,000 vehicle hours of driving.”

Source: Shladover (2017, p. 197)

Technical Challenges in Automating Brown’s driving functions

The following are key technical technological challenges that need to be overcome before automated driving systems will be able to operate safely without constant human supervision (Shladover (2017, p. 198):

• Providing the automated driving system with comprehensive fault detection, identification and accommodation capabilities so that it can immediately diagnose its own malfunctions and switch to a fallback mode of operation that can maintain safety…”.

Source: Shladover (2017, p. 198)

Technical Challenges in Automating Brown’s driving functions

• “Ensuring sufficient cybersecurity protection to repel the large majority of cyber-attacks.”

• “Developing comprehensive environment perception capabilities that can reliably identify, track, and discriminate between benign and hazardous objects in the path of the vehicle under the full range of environmental conditions in which the vehicle is intended to operate (weather and lighting conditions).”

Source: Shladover (2017, p. 198)

Technical Challenges in Automating Brown’s driving functions

• “Resolving questions of “robot ethics” sufficiently to enable the system software to make “life or death” decisions affecting the safety of all road users.”

• “Designing a software-intensive system for a very high level of safety, so that the rate of errors in the system requirements, specifications, and coding is sufficiently low that the system will be no less safe than human driving.”

Source: Shladover (2017, p. 197)

Summary

• Brown’s (1996) driving functions are changing as vehicles become more automated. • The Knowledge, Skills and Behaviours required to operate Automated Vehicles are changing as a result • These changes have implications for Road and Traffic Engineering • There are serious technical challenges that must be overcome in being able to fully automate Brown’s driving functions

Questions ?

Over to ….

Prof. Michael Regan, PhD Research Centre for Integrated Transport Innovation

(rCITI) Room 112, Civil Engineering Building (H20)

E: [email protected]

CVEN4405: Human Factors in Civil and Transport Engineering

Week 10, Lecture 2

Automated and Connected Transport Systems 3: Human

Factors and Traffic Engineering Term 3, 2020

Welcome Back!

79

CVEN 4405 Human Factors in Civil and Transport Engineering

Term 3 2020 Week 10 - Lecture 2

PRESS RECORD BUTTON

+

SHARE SCREEN

80

Lecture Recordings

PLEASE NOTE.

All lectures today are being recorded.

Participation in this meeting indicates your consent to be included in the meeting recording.

Your Course Coordinator and Lecturer

Prof. Michael Regan, PhD Professor of Human Factors

Research Centre for Integrated Transport Innovation (rCITI) School of Civil and Environmental Engineering

University of NSW Sydney

T: +61 (0)2 9385 9504 E: [email protected]

Staff Webpage

82

The CVEN 4405 Teaching Team

Coordinator and Lecturer Prof. Michael Regan Professor of Human Factors Research Centre for Integrated Transport, UNSW Sydney E: [email protected]

Teaching Fellow Dr Prasannah Prabhakharan Research Fellow Research Centre for Integrated Transport, UNSW Sydney E: [email protected]

Demonstrator Mitch Cunningham E:[email protected]. au

Review of Last Lecture

• Impact of Automated and Connected Vehicles on Driving

• Impact of automation on Brown’s driving functions

• The Changing Role of the Driver

• Knowledge, Skills and Behaviours required to use Automated Vehicles

• Implications for Road and Traffic Engineering

• Technical Challenges in Automating Brown’s driving functions

This Lecture - Overview • The Human Factor

• Trust, distrust and over-trust

• Misuse and abuse

• Transition of control

• Skill loss

• Workload

• Education and training

• Loss of Human Cooperation

• Driver Acceptance

• Societal Acceptability

• Ethical Issues

• Motion sickness

• Fatigue, boredom and monotony

• Implications for road and traffic engineering

85

Learning Outcomes

CLO1: Explain the fundamental principles of HF that can be used by civil and transport engineers to facilitate user- centred design

CLO2: Apply HF principles, methods and data to the design of road and traffic management systems

CLO3: Plan for the integration of HF into the design lifecycle of the road and traffic management system

The Human Factor – Automated Vehicles

87

The Human Factor – Automated Vehicles

Vehicle automation is not yet 100% reliable in performing all of Brown’s driving activities

The human driver still needed at SAE Level 3 to resume manual control if the autonomous system fails or reaches its limits of competence

This creates some humans factors issues that must be addressed as we progress towards SAE Level 5 vehicles. These issues are the focus of this lecture.

88

Issues & Challenges: Trust

Human trust in automation is itself not entirely calibrated

Sometimes it is too low (distrust), and sometimes too high (overtrust)

Source: Wickens, Lee et al., (2004); Fisher et al. (2020)

89

Issues & Challenges: Trust – Use Cases

The following are five key use cases that Weast et al. (2016) argue will be critical to address in building trust in autonomous vehicles (e.g. “robotaxis” with no driver):

• Requesting an AV • Entering an AV and initiating a trip • Making trip changes (intended or unintended) in an AV • Safely pulling over and exiting an AV • Using the road in proximity to an AV

Source: Weast et al. (2016); Fisher et al. (2020)

90

Issues & Challenges: Distrust

Occurs when the human fails to trust automation as much as is appropriate

Has been shown to compromise safety • e.g. Driver takes back manual control of vehicle, even though the vehicle

is more capable of avoiding a collision than the driver. (has been a major problem in commercial aviation).

• e.g. Driver doesn’t utilise the technology at all, negating all safety and other benefits

Source: Wickens, Lee et al. (2004); Parasuraman & Riley (1997); Fisher et al. (2020)

91

Issues & Challenges: Overtrust

Overtrust, or “complacency”, occurs when humans trust automation more than is warranted

Overtrust results from the human tendency to let experience guide their expectancies.

If humans never encounter a system failure, they may perceive its reliability to be perfect, even if it is not… can be problematic

• e.g. Truck driver is not confident that the platooning system is functioning properly, but nevertheless decides not to take back control of the vehicle – HUGE PROBLEM in AVIATION!

Source: Wickens, Lee et al. (2004); Parasuraman & Riley (1997); Fisher et al. (2020).

92

Issues & Challenges: Overtrust (2)

Complacency may lead to humans failing to monitor systems adequately, and as a result getting themselves “out of the loop”

This is problematic if automation fails and the driver must intervene to take back control

Complacent drivers – for example in Level 3 vehicles – who are out of the loop will be likely to:

• be slower to detect an automation failure • be less likely to intervene correctly • and may lack the manual skills required to intervene, because of having

relied too much on automation

Source: Wickens, Lee et al. (2004); Parasuraman & Riley (1997); Fisher et al. (2020).

93

Issues & Challenges: Misuse and Abuse

Misuse - a lack of understanding of the capabilities and limitations of vehicle automation may result in drivers falsely assuming that the vehicle is more capable than it is e.g. reverse parking aid

Abuse - inappropriately high levels of trust in a system might encourage drivers to deliberately use the system beyond its operational envelope or operational design domain. e.g. recent Tesla crash.

Source: Parasuraman & Riley (1997); Fisher et al. (2020)

94

Issues & Challenges: Transition of Control

In Level 3 vehicles, driver must be available to resume control if requested by the vehicle automation

Challenges: • How to keep driver in the loop and ready to intervene if necessary. What

if driver is fatigued, sleepy or distracted?

• How to support driver to take back control in a way that reduces take over time and maximises take over quality and safety?

May be less of a problem in trucking industry with well trained, professional and regulated drivers.

Source: Classen & Alvarez (2020); Fisher et al. (2020)

95

Issues & Challenges: Skill Loss

Gradual loss of skill will result if drivers have not been in control of the system for prolonged periods of time, because of automation.

Loss of skill has been shown to: • make humans less confident in their own performance and hence more

likely to continue to use automation • degrade their ability to intervene if required to take back control

Source: Lee & Moray, (1994); Parasuraman et al., (2000); Debus, Wang, & Huestegge, (2014); Fisher et al. (2020)

96

Issues & Challenges: Skill Loss (2)

It may also degrade driver ability to drive manual vehicles if required

• Even quite brief periods of automation can degrade manual driving skills • Impact of skill loss on drivers of AVs is currently unknown.

Source: Lee & Moray, (1994); Parasuraman et al., (2000); Debus, Wang, & Huestegge, (2014); Fisher et al. (2020)

97

Issues & Challenges: Workload

Automation may be problematic if it: • reduces workload during already low-workload periods, leading to loss

of arousal eg ACC on highways: – Which may induce passive fatigue  reduced attentional resources 

increased reaction time to unexpected events

– Which may also encourage drivers to engage in secondary tasks - which has been shown to adversely affect take-over time and quality.

• increases workload during high workload periods e.g. pilots programming FMS during landing preparation

Source: Neubauer et al., (2012); Merat et al., (2012); Fisher et al. (2020)

98

Issues & Challenges: Education and Training

As vehicles become increasingly automated, the roles of the driver, as noted earlier, will change.

With these changing roles, will come a change in the knowledge, skills and behaviours that are required to safely operate automated vehicles from Levels 1 to 3.

Source: Regan, Prabhakharan et al. (2019); Fisher et al. (2020)

99

Issues & Challenges: Education and Training (Knowledge)

Drivers of partially automated vehicles (Level 1-3), will require new knowledge for safe operation.

For example: • Environmental conditions that may degrade system performance • Environmental conditions that are unsuitable for the use of system • System performance limitations not related to driving conditions,

including causes of degraded system performance • Symptoms of system malfunctions

Source: Regan, Prabhakharan et al. (2019); Fisher et al. (2020)

100

Issues & Challenges: Education and Training (Skills)

Drivers of partially automated vehicles (Level 1-3), will require new skills for safe operation.

For example: • Recognizing environmental conditions that may cause system

performance to be ineffective or degraded • Recognizing environmental conditions that may degrade system

performance • Distinguishing between normal and aberrant system performance

Source: Regan, Prabhakharan et al. (2019); Fisher et al. (2020)

101

Issues & Challenges: Education and Training (Behaviour)

Drivers of partially automated vehicles (Level 1-3), will require new behaviours for safe operation.

For example: • Maintaining vigilance of the driving environment, system operating

modes and associated vehicle performance • Willingness to use system when appropriate to do so • Avoidance of the use of the system when inappropriate to do so.

Source: Regan, Prabhakharan et al. (2019); Fisher et al. (2020)

102

Issues & Challenges: Human Cooperation

In non-automated vehicles, drivers cooperate with each other to signal their intentions and avoid crashes.

It is important to ensure that highly automated and connected vehicles do not overlook important information and communication channels that exist presently to eliminate traffic conflicts and crashes.

Source: Wickens, Lee et al. (2004); Fisher et al. (2020)

103

Issues & Challenges: Driver Acceptance

If automation is unacceptable to drivers (not perceived to be reliable, useful or satisfying to use), they may refuse to use it, or misuse or abuse it - and it will have no safety or other benefit.

Not all drivers will want driving tasks to be replaced by automation.

Source: Wickens, Lee et al. (2004); Fisher et al. (2020)

104

Issues & Challenges: Driver Acceptance (2)

How to deal with that will be a challenge: • Should vehicle manufacturers give drivers the option of driving self-

driving vehicles manually when they want?

• Should drivers be banned from driving manual vehicles when all vehicles are self-driving?

• Should drivers be allowed to drive vehicles manually only on closed circuits?

Source: Wickens, Lee et al. (2004); Fisher et al. (2020)

105

Issues & Challenges: Societal Acceptability

What do Australians and New Zealanders think about automated, autonomous and connected vehicles?

Survey Working Group of Australian Driverless Vehicle Initiative (ADVI) – created by ARRB with 120 members

Designed and administered 2 national surveys of driver awareness, opinions and likely acceptance of automated and driverless vehicles

• 2016 – 5000+ Australian respondents • 2017 – 6,100+ Australians and New Zealanders

Source: Cunningham et al., (2019)

106

Issues & Challenges: Societal Acceptability (ADVI Survey 1)

Most Australians are aware of most automated vehicles functions, but very few have experienced them

There is high-level of community concern about many issues relating to completely self-driving cars

Less than half of all people are willing to for self-driving vehicles than for their existing car

Source: Cunningham et al., (2019)

107

Issues & Challenges: Societal Acceptability (ADVI Survey 1) (2)

Most agree that there are many potential benefits from AVs

Most people are comfortable with driverless vehicles controlling most driving functions

People are least comfortable with cars changing lanes by themselves and following cars ahead too closely

Source: Cunningham et al., (2019)

108

Issues & Challenges: Societal Acceptability (ADVI Survey 1) (3)

Most people are likely to spend their time observing scenery and interacting with passengers in fully self-driving cars

People are more comfortable about taking control than giving control to semi-automated vehicles

Most people would like to drive a fully self-driving car manually, from time to time

Source: Cunningham et al., (2019)

109

Issues & Challenges: Societal Acceptability (ADVI Survey 1) (4)

Less than half of people think that driverless cars could be safer than a car driven manually by a human

Females and males think differently about AVs

People in different states and territories think differently about AVs

Source: Cunningham et al., (2019)

110

Issues & Challenges: Ethical Issues

How to design a decision algorithm in the face of an inevitable crash?

Bonnefon (2016) posed alternative algorithms to an imminent crash to a group of participants.

• e.g. Driver approaching pedestrians. Drive into pedestrians and kill them? Or veer off road and get killed?

Source: Bonnefon (2016)

111

Issues & Challenges: Ethical Issues (2)

Most respondents opted for utilitarian design option - i.e. to minimise no. of injuries.

But if they bought a car, would they want it programmed to minimise total injuries or protect its occupant(s) at all costs?

Most preferred the latter, and disapproved of governments mandating the utilitarian option.

Source: Bonnefon (2016)

112

Issues & Challenges: Motion Sickness

Occurs when movement is out of control of humans and there is reduced ability to anticipate movement.

Up to 10% of American adults are expected to experience motion sickness in highly automated vehicles.

May reduce driver acceptance and utilisation of automation. There is some evidence that some population may be more

prone to motion sickness in automated vehicles than others – E.g. in terms of Ethnicity and Gender

Source: Sivak & Schoettle, (2015); Klosterhalfen et al. (2005).

113

Issues & Challenges: Fatigue, Boredom & Monotony

Boredom and monotony are widely recognised as foreseeable side effects of vehicle automation.

• Recall the inverse U-shaped curve of arousal? • Engagement in the driving task maintain arousal in the

optimal zone.

Drivers in automated vehicles may be prone to become fatigued faster than manual drivers.

Source: Schömig et al., (2015), Vogelpohla et al (2019)

The Human Factor – Connected Automated Vehicles

115

The Human Factor - Connected Automated Vehicles • We have discussed a number HF issues relating to

automated vehicles • When automated vehicles become connected, it is hard to

know at this point in time whether these HF issues will change: e.g.

• If vehicles are more capable of driving like humans, will drivers may have more trust in them?

• If they are more capable of driving like humans, will drivers abuse them less or more?

• Will connectivity be able to alert drivers in advance to situations that require them to take back control, so that they are more ready to do so?

HF Issues - Implications for Road and Traffic Engineers

• Road and traffic engineers can have some influence over these HF issues generated by automation and connectivity:

• e.g. they can implement treatments, discussed earlier in the course, to prevent drivers from crashing due to driver inattention and distraction in Level 2 and 3 vehicles (e.g. rumble strips)

• e.g. they can, as discussed earlier in the course, design roads to be forgiving in the event that drivers lose control or crash as a result of skill loss, abuse, transfer of control difficulties, distraction, etc

117

Summary

• Automated vehicles and connected automated vehicles have significant potential to improve traffic safety and mobility

• The actual benefits are unknown, as we cannot predict with certainty how they will roll out

• The human factor will be critical for the success of these technologies, even when all vehicles are driverless

• Even then, humans will be responsible for programming them to think and behave

• Road and traffic engineers have some important influence over the HF issues generated by automation and connectivity

118

Reflections…

Thanks From all of Us !

MIKE PRASANNAH MITCH We enjoyed teaching you and we hope you enjoyed the

course! Remember to put the human as the centre of attention in

system development

Good luck with your careers!

Questions ?

Over to ….

121

Thursday Tutorial

Are you all prepared for your group presentations? I’ll be there with Mitch

Any questions?

Prof. Michael Regan, PhD Research Centre for Integrated Transport Innovation

(rCITI) Room 112, Civil Engineering Building (H20)

E: [email protected]